...
Attendees: Patrick, Stephanie, Till, Diane, Debbie, Tom Himmel, Hamid Shoaee
A first draft of the Fast Feedback Requirements document was emailed with the meeting announcement.
It is uploaded to Sharepoint under Controls | Fast Feedback | Beam Based Fast Feedback Requirements
Review of some Assumptions/Requirements for the Fast Feedback ProjectThe Project
- The controller application will be configurable to work with either Channel Access or Fast Feedback network. Slow loops may simply use CA.
- The actuators must also be configurable to work with either Channel Access control, or Fast Feedback network control
- measurement IOCs will work on both CA and FF networks, BSA records will be used by feedback loops that are using the CA network.
- individual feedback loops will be configured by time-slot. 120Hz operation will be achieved by assigning two independent loops operating on timeslots 1 and 4 at 60 Hz each.
- The beam actuator devices need a new EPICS design of a Pulsed Amplitude Unit (PAU). The intention is incorporate this into the existing magnet and PAC IOCs. The PAU design would be a front end with six independent user inputs for DES values, each one associated with one time slot. The six inputs are multiplexed to a single output at 360 Hz. This 360 Hz waveform would be the DES input for an existing magnet or PAC controller. Similarly, the ACT readback would be a 360 Hz waveform which would be unmultiplexed into 6 channels.
Design Discussions
PAU DiscussionsPAU
The last assumption above was the lead-in to a lively discussion on PAU design. The main questions under discussion:
- Is 6 inputs and pathways really necessary?
The actuators themselves will never operate any faster than 120Hz, requiring only two inputs (for slots 1 and 4). Stephanie is concerned that the 360Hz operation of the PAU will required alot of overhead in processing, reducing the available processing power already needed by the actuator IOCs.
- When are the actuator control inputs applied?
We drew up a timeline of finducials with beam on timeslots 1, and 4. We concluded that the input for timeslots 1,5,6 must be applied after fiducial 2(timeslot 2), inputs for timeslots 2,3,4 must be applied after fiducial 5 (timeslot 5). This drawing will be added to the requirements (and yet to be created design) documents.\
EVENT CODES
It was agreed that the Event Codes assign for Feedback will be per timeslot. A delay from this event code trigger will be implemented by the controller application using a hardware timer on the VME6100. Note that this is processor dependent and will require 'porting' if we move our IOCs to a new processor. Till already has a timer driver and can provide a framework that will allow the programmer to implement multiple delays (in case of more than one controller on an IOC). This framework will work with Linux as well, so debugging on Linux will be possible.
Action Items
- Patrick volunteered to write up details about this PAU .
- Diane will document the timeline drawing on the board
- Diane to continue refining Requirements Document
- Till to provide timer driver and framework